Abstract
In this article, we establish the unique global solvability of a 2D stochastic Cahn–Hilliard–Oldroyd model of order one, for the motion of an incompressible isothermal mixture of two (partially) immiscible non-Newtonian fluids having the same density and perturbed by a multiplicative noise of Gaussian and Lévy type. The model consists of the stochastic Oldroyd model of order one, coupled with a stochastic Cahn–Hilliard model. We prove the existence and uniqueness of a strong solution (in the stochastic sense). The proofs are based on the Galerkin approximation technique. Moreover, we also prove that under some conditions on the forcing terms, the strong solution converges exponentially in the mean square and almost surely exponentially to the stationary solutions.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
The Oldroyd model of order one arises in the dynamics of non-Newtonian fluids and is well-known as the generalization of the initial-boundary value problem for the Navier–Stokes equations. It is used to model the motion of viscoelastic fluids (see [23, 31]). The analysis of this model, both in the deterministic and the stochastic cases has been investigated by several authors (see for instance [23, 25,26,27, 31, 32]). The well-posedness and the exponential stability of the model in two-dimensional bounded and unbounded (Poincaré domains) domains, both in deterministic and stochastic settings is studied in [27]. The proof of the existence and the uniqueness of weak solution in the deterministic case is obtained via a local monotonicity property of the linear and nonlinear operators and a localized version of the Minty-Browder technique. The global solvability results for the stochastic counterpart are obtained by a stochastic generalization of the Minty-Browder technique. In order to describe the behavior of complex fluids in Fluid Mechanics, diffuse-interface methods are widely used by many researchers (see, e.g., [3, 15] and references therein). A typical example is a mixture of two incompressible fluids. The evolution of such a system is described by a sufficiently simple model so-called H model (see [19, 20, 30] and references therein). This consists in a suitable coupling of the Navier–Stokes equations for the (average) fluid velocity u through a capillarity force proportional to \(\mu \nabla \phi \), where \(\mu \) is the chemical potential, with a local or nonlocal Cahn–Hilliard type equation for the order parameter \(\phi \) through a transport term \(u\cdot \nabla \phi \). The mathematical and the numerical analysis of the deterministic and the stochastic local and nonlocal Cahn–Hilliard–Navier–Stokes (CH-NS) model has been considered by several authors such as [9,10,11,12,13,14, 17, 18, 21, 22, 24, 36,37,38] and references therein. In [13], the authors considered the stochastic 3D globally modified local CH–NS equations with multiplicative Gaussian noise. They proved the existence and uniqueness of strong solution (in the sense of partial differential equations and stochastic analysis) and derived the existence of a weak martingale solution for the stochastic 3D local CH-NS equations. The third author of the paper has proved the existence and uniqueness of the probabilistic strong solution for the stochastic 2D local CH-NS model with multiplicative Gaussian type of noise in [36] and studies the asymptotic stability of the unique probabilistic strong solution of 2D local CH-NS model in [37]. He also proved the existence of weak solution of 3D local CH-NS model with multiplicative non-Gaussian Lévy noise in [38]. Similar results have been obtained for the 2D and 3D nonlocal CH-NS model in [9, 11, 12]. Generally, the CH-NS model is used to modelise the flow of a Newtonian binary mixture. However, for viscoelastic binary fluids, the behavior of viscoelastic fluids cannot be predicted by the means of usual Newton’s constitutive law since they possess a memory of past deformations which is not the case for Newtonian fluids. Consequently, we have to introduce a more general phenomenological model such as Oldroyd model in addition to the Navier–Stokes equation as a constitutive equation to modelise the viscoelasticity. Taking into account this fact, we derive a model where we will call Cahn–Hilliard–Oldroyd model.
In this article, we study a stochastic generalization model of the CH-NS model. More precisely, we consider in a complete probability space \((\Omega ,{\mathcal {F}},{\mathbb {P}})\) equipped with an increasing family of sub-sigma fields \(\{{\mathcal {F}}_t\}_{0\le t\le T}\) of \({\mathcal {F}}\) such that \({\mathcal {F}}_0\) contains all elements \(F \in {\mathcal {F}}\) with \({\mathbb {P}}(F) = 0\) and \({\mathcal {F}}_t= \cap _{s>t}{\mathcal {F}}_s\) for \(0 \le t \le T\), the following stochastic Cahn–Hilliard–Oldroyd model of order one, for the motion of an incompressible isothermal mixture of two immiscible non-Newtonian fluids
in \((0,T)\times {\mathcal {M}}\) with the conditions
where \(T>0\), \({\mathcal {M}}\) is a bounded open domain in \({\mathbb {R}}^2\) with a smooth boundary \(\partial {\mathcal {M}}\) and
In (1.1), \(W^i_t\), \(i=1,2\), are two cylindrical Wiener processes in a separable Hilbert space U defined on the probability space \((\Omega ,{\mathcal {F}},{\mathbb {P}})\). Also Z is a measurable subspace of some Hilbert space and \({\tilde{\pi }}(dt, dz):= \pi (dt, dz) - \lambda (dz)dt\) is a compensated Poisson random measure, where \(\lambda (dz)\) is a \(\sigma \)-finite Lévy measure on the Hilbert space with an associated Poisson random measure \(\pi (dt, dz)\) such that \({\mathbb {E}}[\pi (dt, dz)] = \lambda (dz)dt\). The processes \(W^i_t\), \(i=1,2\), and \({\tilde{\pi }}\) are mutually independent. Let \(\nu > 0\) be the coefficient of kinematic viscosity. In (1.1), for \(\varsigma ,\kappa > 0,\) we take
If we take \(\nu _1=\frac{\kappa }{\varsigma }\) (or \(\gamma =0\)) in (1.3), then we obtain a model for the phase separation of an incompressible and isothermal Newtonian binary fluid flow or the well-known Cahn–Hilliard–Navier–Stokes model [13, 37, 38]. Note that \(\varsigma \) is the relaxation time, and is characterized by the fact that after instantaneous cessation of motion, the stresses in the fluid do not vanish instantaneously, but die out like \(e^{-\varsigma ^{-t}}\). Moreover, the velocities of the flow, after instantaneous removal of the stresses, die out like \(e^{-\kappa ^{-t}}\), where \(\kappa \) is the retardation time. For the physical background and the mathematical modeling of viscoelastic fluid flows involving memory effects, we refer the interested readers to [2, 23, 34], where the topic is studied extensively. The quantity \(\mu \) is the chemical potential of the binary mixture which is given by the variational derivative of the following free energy functional
where, \(F(r)=\int _0^rf(\zeta )d\zeta \) is the suitable double-well potential. The quantities \(\nu _2\) and \({\mathcal {K}}\) are positive constants that correspond to the mobility constant and capillarity (stress) coefficient, respectively. \(\varepsilon \) and \(\alpha \) are two positive parameters describing the interactions between the two phases. In particular, \(\varepsilon \) is related to the thickness of the interface separating the two fluids. A typical example of potential F is of logarithmic type. However, this potential is very often replaced by a polynomial approximation of the type \(F(r)=\gamma _1r^4-\gamma _2r^2,\) with \(\gamma _1\) and \(\gamma _2\) are positive constants. Note that in (1.1), the \(\phi \) equation modelises the evolution of the concentration of fluids which can be influenced by the thermal fluctuations which is a random phenomena. In order to take into account this thermal fluctuations, we have introduce a noise in addition to the \(\phi \) equation as a constitutive equation. However, a Levy type noise in the \(\phi \) equation is also possible but, the presence of such noise will involve probably more assumptions and will increase significantly the size of the paper.
To the best of our knowledge, there is no mathematical analysis of the model (1.1) even in its deterministic setting. This article is a contribution in that direction. Moreover, in the literature, there is no work on stochastic two-phase flows models with both Gaussian and non-Gaussian type of noise. But a such type of noise have been considered for instance in [27, 29] for the Oldroyd model of order one and the Navier–Stokes equation with hereditary viscosity. In general the presence of a noise on the concentration equation in the two-phase flow model makes the analysis of the model more involved (see [13, 16, 37]). In [13], an existence result has been obtained under the additional strong condition on the potential f. In order to use a weakened condition on the potential function f for the existence result, we have shown that the energy functional \({\mathcal {E}}_0\) is twice Fréchet differentiable which makes possible the application of Itô’s formula to the process \({\mathcal {E}}_0(\phi )\). Note that in (1.1) it is possible to add a Lévy type of noise on the relative concentration equation, but the analysis will be tedious. The purpose of the present manuscript is to prove some results related to problem (1.1). Our main results are the following: First, we prove the existence and uniqueness of strong solution (in the stochastic sense) for system (1.1). The method for the proof is based in the Galerkin approximation. Secondly, getting the existence of a unique strong solution in hand, we investigate the stability of this solution. More precisely, we prove that under some conditions on the forcing terms, the strong solution converges exponentially in the mean square and almost surely exponentially to the stationary solutions.
The rest of the paper is organized as follows. In the next section, we describe the mathematical setting required to establish the unique solvability of the system (1.1). The hypothesis satisfied by the potential, the noise coefficient and the external forcing are also discussed in the section. In Sect. 3, we introduce the Galerkin approximation of our problem and we derive a priori estimates for its solution. Then we prove the existence and the pathwise uniqueness of strong solution. In the last section, we analyze the stability of stationary solutions.
2 Functional Setting, Hypothesis and Abstract Formulation
In this section, we fix the hypothesis and describe the functional spaces needed to establish the existence and uniqueness of global strong solution of the system (1.1). We discuss the properties of linear and nonlinear operators, and also of the kernel \(\beta (t) =\gamma e^{-\delta t}\).
2.1 Functional Setting
We now introduce the functional setup of Eqs. (1.1)–(1.2). If X is real Hilbert space with inner product \((.,.)_{X}\), then we denote the induced norm by \(|.|_{X}\), while \(X^*\) will indicate its dual. Let us consider the Hilbert spaces
where \({\mathbb {L}}^2({\mathcal {M}}):=(L^2({\mathcal {M}}))^2\), \({\mathbb {H}}_{0}^1({\mathcal {M}}):=(H_{0}^1({\mathcal {M}}))^2\). We endow \(H_{1}\) with the \(L^2-\)inner product and norm
Moreover, the space \(V_{1}\) is endowed with the scalar product and norm
The norm in \(V_{1}\) is equivalent to the \({\mathbb {H}}^1({\mathcal {M}})\)-norm (due to Poincaré’s inequality). We refer the reader to [35] for more details on these spaces.
We now define the operator \(A_{0}\) by
where \({\mathcal {P}}\) is the Leray-Helmholtz projector in \({\mathbb {L}}^2({\mathcal {M}})\) onto \(H_{1}\). Then, \(A_{0}\) is a self-adjoint positive unbounded in \(H_{1}\) which is associated with the scalar product defined above. Furthermore, \(A_{0}^{-1}\) is a compact linear operator on \(H_{1}\) and by the classical spectral theorem, there exists a sequence \(\lambda _{j}\) with \(0<\lambda _1<\lambda _2\le \cdots \le \lambda _n\le \lambda _{n+1}\le \cdots \) and a family \(w_j\in D(A_0)\) which is an orthonormal basis in \(H_1\) and such that \(A_0w_j=\lambda _jw_j\).
For \(u\in H_1\), we denote \(u_j=(u,w_j)\). Given \(\alpha >0\), take
and define \(A_0^{\alpha }u=\sum _j\lambda _j^\alpha u_j w_j\), \(u\in D(A_0^\alpha )\). We equip \( D(A_0^\alpha )\) with the norm \(|u|_{\alpha }^2:=|A_0^\alpha u|_{L^2}^2=\sum _j\lambda _j^{2\alpha }|u_j|^2\).
We introduce the linear nonnegative unbounded operator on \(L^2({\mathcal {M}})\)
and we endow \(D(A_{1})\) with the norm \(|A_{1}\cdot |+|\left\langle \cdot \right\rangle |\), which is equivalent to the \(H^2\)-norm. We also define the linear positive unbounded operator on the Hilbert space \(L_{0}^2({\mathcal {M}})\) of the \(L^2\)-functions with null mean
Note that \(B_{n}^{-1}\) is a compact linear operator on \(L_{0}^2({\mathcal {M}})\). More generally, we can define \(B_{n}^s\), for any \(s\in {\mathbb {R}}\), noting that \(|B_{n}^{s/2}\cdot |_{L^2}\), \(s>0\), is an equivalent norm to the canonical \(H^s\)-norm on \(D(B_{n}^{s/2})\subset H^s({\mathcal {M}})\cap L_{0}^2({\mathcal {M}})\). Also note that \(A_{1}=B_{n}\) on \(D(B_{n})\). If \(\varphi \) is such that \(\varphi -\left\langle \varphi \right\rangle \in D(B_{n}^{s/2})\), we have that \(|B_{n}^{s/2}(\varphi -\left\langle \varphi \right\rangle )|_{L^2}+|\varphi -\left\langle \varphi \right\rangle |_{L^2}\) is equivalent to the \(H^s\)-norm. Moreover, we set \(H^{-s}({\mathcal {M}})=(H^s({\mathcal {M}}))'\), whenever \(s<0\).
We note that
Classically, there exists a sequence \(\beta _{j}\) with \(0<\beta _1<\beta _2\le \cdots \le \beta _n\le \beta _{n+1}\le \cdots \) and a family \(\psi _j\in D(A_1)\) which is an orthonormal basis in \(L^2({\mathcal {M}})\) and such that \(A_1\psi _j=\beta _j\psi _j\).
Now for \(\alpha \ge 0\) we define
endowed with the Hilbertian norm \(|\phi |_{\alpha }^2:=|A_1^\alpha \phi |_{L^2}^2=\sum _j\beta _j^{2\alpha }|(\phi ,\psi _j)|^2\).
We introduce the bilinear operators \(B_{0}, B_{1}\) (and their associated trilinear forms \(b_{0}, b_{1}\)) as well as the coupling mapping \(R_{0}\) which are defined, from \(D(A_{0})\times D(A_{0})\) into \(H_{1}\), \(D(A_{0})\times D(A_{1})\) into \(L^2({\mathcal {M}})\) and \(L^2({\mathcal {M}})\times D(A_{1}^{3/2})\) into \(H_{1}\), respectively. More precisely, we set
Note that
Using an integration by part, we can check that
We recall from [18] that, using the integration by parts, a suitable generalized Hölder inequality and a suitable Gagliardo-Nirenberg interpolation inequality, we derive that \(b_0,\) and \(b_1\) satisfy the following properties.
owing to (1.2)\(_1\) we derive that
From (2.13), we deduce the mass conservation in he deterministic case. In fact, for \(\sigma _2(\cdot )=0\), from (2.13), we have the conservation of the following quantity
where \(\left| {\mathcal {M}}\right| \) stands for the Lebesgue measure of the domain \({\mathcal {M}}\). More precisely, we have
Hereafter, we assume that \(\sigma _2(\cdot )\) is chosen such that (2.14) is satisfied, which is the case if we assume that
where \(H_2\) is defined by (2.19) below.
Due to the mass conservation, we have
Thus, up to a shift of the order parameter field, we can always assume that the mean of \(\phi \) is zero a the initial time and, therefore it will remain zero for all positive times. Hereafter, we assume that
We set
The norm in \(H_2\) is denoted by \(\Vert \cdot \Vert \), where \(\Vert \psi \Vert ^2=|A_1^{1/2}\psi |_{L^2}^2\). The space \({\mathcal {H}}\) is a complete metric space with respect to the metric associated with the norm
We set \(V_2=D(A_1)\) and define the Hilbert spaces \({\mathcal {U}}\) and \({\mathbb {V}}\) respectively by
endowed with the scalar product whose associated norm are respectively
We will denote by \(\lambda _0\) a positive constant such that
For \(u_1=(v_1,\phi _1),\) \(u_2=(v_2,\phi _2),\) \(u_3=(v_3,\phi _3)\in {\mathbb {V}},\) we define
Lemma 2.1
The maps B, R and E are locally Lipschitz continuous i.e. for every \(r>0,\) there exists a constant \(C_r\) such that
for all \(v_1=(u_1,\phi _1),\) \(u_2=(u_2,\phi _2)\in {\mathbb {V}}\) with \(\left\| v_1\right\| _{\mathbb {V}}\) and \(\left\| v_2\right\| _{\mathbb {V}}\le r\), where \({\mathbb {V}}^{*}\) is the dual space of \({\mathbb {V}}\).
Proof
Let \(v_1=(u_1,\phi _1),\) \(u_2=(u_2,\phi _2)\in {\mathbb {V}}\) and \((w,\psi )\in {\mathbb {V}}.\) We assume that \(\left\| v_1\right\| _{\mathbb {V}}\le r\) and \(\left\| v_2\right\| _{\mathbb {V}}\le r.\) To prove (2.24)\(_1\), we note that
Also,
which implies that
Proceeding similarly as in (2.25) and using the fact that \(D(A_1)\) is continuously embedded in \(L^2({\mathcal {M}})\), we obtain
It then follows that
From (2.25) and (2.26) we derive that
which prove (2.24)\(_1.\) Now, remark that
However,
It follows that
which prove (2.24)\(_2.\) Also, we remark that
Let us recall from [18] that, there exists a monotone non-decreasing function \(Q_1(x_1,x_2)\) of \(x_1\) and \(x_2\) such that
We recall that, since \(\left\| \phi _1\right\| \le c\left| A_1\phi _1\right| _{L^2}\le cr,\) \(\left\| \phi _2\right\| \le c\left| A_1\phi _2\right| _{L^2}\le cr\) and \(Q_1\) is a monotone non-decreasing function, we have \(Q_1(\left\| \phi _1\right\| ,\left\| \phi _2\right\| )\le Q_1(cr,cr).\) So, we obtain
Hence
which proves (2.24)\(_3\) and ends the proof of Lemma 2.1. \(\square \)
Now, we discuss some properties of a general kernel \(\beta (.)\) and then in particular \(\beta (t)=\gamma e^{-\delta t}\). we define
A function \(\beta (\cdot )\) is called positive kernel if the operator L is positive on \(L^2(0,T;H_1)\) for all \(T>0\). That is,
for all \(u\in L^2(0,T;H_1)\) and every \(T>0\).
Let \({\hat{\beta }}(\theta )\) be the Laplace transform of \(\beta (t)\), i.e.
Then we have from [5, Lemma 4.1] the following result.
Lemma 2.2
Let \(\beta \in L^\infty (0,\infty )\) be such that \(Re {\hat{\beta }}(\theta )>0\) if \(Re(\theta )>0\). Then, \(\beta (t)\) defines a positive kernel.
We also recall from [29, Lemma A\(_1\)] the following result.
Lemma 2.3
Let \(\beta (t)=\gamma e^{-\delta t}\), \(\delta >0\), \(t\in [0,T]\). Then for any right continuous function with left limits, \(f:[0,T]\rightarrow [0,\infty )\), we have
for all \(u\in L^2(0,T; H_1)\).
Remark 2.1
-
(1)
As proved in [28, Lemma 2.6], with a change of variable and change of integrals, it can be easily seen that, if \(\beta \in L^1(0,T)\), \(f,g\in L^2(0,T)\), for some \(T>0\), then we have
$$\begin{aligned} \left( \int _0^Tg^2(t)dt\left( \int _0^t\beta (t-s)f(s)ds\right) ^2\right) ^{1/2}\le \left( \int _0^T\left| \beta (t)\right| dt\right) \left( \int _0^Tg^2(t)f^2(t)dt\right) ^{1/2}. \end{aligned}$$(2.28) -
(2)
If we take \(g(t)=1\), and \(f(t)=\left| u(t)\right| _{L^2}\) in (2.28) with \(u\in L^2(0,T;H_1)\), we obtain
$$\begin{aligned} \left( \int _0^T\left( \int _0^t\beta (t-s)\left| u(s)\right| ^2_{L^2}ds\right) ^2\right) ^{1/2}\le \left( \int _0^T\left| \beta (t)\right| dt\right) \left( \int _0^T\left| u(t)\right| ^2_{L^2}dt\right) ^{1/2}. \end{aligned}$$(2.29) -
(3)
For \(\beta (t)=\gamma e^{-\delta t}\), we know that
$$\begin{aligned} \int _0^\infty \beta (t)dt=\frac{\gamma }{\delta }\ \text { and } \ {\hat{\beta }}(\theta )=\frac{\gamma }{\theta +\delta }>0,\ \text { for } \ Re\theta >0, \end{aligned}$$and by Lemma 2.2, \(\beta (t)\) is a positive kernel. Hence, we have
$$\begin{aligned} \int _0^T\langle (\beta *A_0u)(t), u(t) \rangle dt=\int _0^T( (\beta *\nabla u)(t),\nabla u(t) )dt\ge 0. \end{aligned}$$(2.30)Using the fact that \(\left\| A_0u\right\| _{V_1^*} \le \left\| u\right\| \), we get
$$\begin{aligned} \left\| (\beta *A_0u)(t)\right\| _{V_1^*}\le \int _0^t\beta (t-s)\left\| A_0u(s)\right\| _{V_1^*}ds\le \int _0^t\beta (t-s)\left\| u(s)\right\| ds, \end{aligned}$$(2.31)and hence by (2.29), we have
$$\begin{aligned} \int _0^T\left\| (\beta *A_0u)(t)\right\| ^2_{V_1^*}dt&\le \int _0^T\left( \int _0^t\beta (t-s)\left\| A_0u(s)\right\| _{V_1^*}ds\right) ^2dt\nonumber \\&\le \left( \int _0^T\beta (t)dt\right) ^2\int _0^T\left\| u(t)\right\| ^2dt\le \frac{\gamma ^2}{\delta ^2}\int _0^T\left\| u(t)\right\| ^2dt, \end{aligned}$$(2.32)for \(u\in L^2(0,T;V_1)\).
Remark 2.2
Using the Cauchy-Schwarz and Hölder inequalities and (2.32), we derive that
2.2 Hypothesis
We assume that \(W^i\), \(i = 1,2\) are formally given by the expansion
where \(\beta _j(t)\), \(j \in {\mathbb {N}}\) are independent one dimensional Brownian motions on \((\Omega , {\mathcal {F}}, {\mathbb {P}})\), and \(\{ \beta _j\}_{j=1}^\infty \) is an orthonormal basis on U. We also define the auxiliary space \(U_0\) containing U, that is defined by
endowed with the scalar product
The stochastic forcing takes the following form
with suitable restrictions on the growth of the diffusion coefficients \(\sigma _j^i\) specified below.
Let us denote by \({\mathbf {D}}([0, T ]; H_1)\), the set of all \(H_1\)-valued functions defined on [0, T], which are right continuous and have left limits (Càdlàg functions) for every \(t \in [0, T ]\). Also, let
be the space of all \({\mathcal {B}}((0,T]\times {\mathcal {F}}\times Z )\) measurable functions \(\gamma :[0,T]\times \Omega \times Z\rightarrow H_1\) such that
For any Hilbert space H, we will denote by \({\mathcal {L}}^2(U;H)\) the separable Hilbert space of Hilbert-Schmidt operators from U into H.
To simplify the notations, we set (without loss of generality) \(\nu _1 = \nu _2 = \varepsilon = \alpha = {\mathcal {K}}= 1\). Let us assume that the potential function f and the noise coefficients \(\sigma _i(\cdot , \cdot )\), \(i=1,2\) and \(\gamma (\cdot , \cdot , \cdot )\) satisfy the following hypothesis.
- (H1):
-
\(f\in {\mathcal {C}}^{2}({\mathbb {R}})\) satisfies
$$\begin{aligned} {\left\{ \begin{array}{ll} \liminf _{|r|\rightarrow +\infty } f'(r)>0, \\ |f^{(i)}(r)|\le c_{f}(1+|r|^{2-i}),~\forall r\in {\mathbb {R}},~i=0,1,2, \end{array}\right. } \end{aligned}$$(2.37)where \(c_{f}\) is some positive constant.
- (H2):
-
For all \(t\in [0,T]\), \(\langle \int _0^t\sigma _2(s,u,\phi )dW^2_s\rangle =0\) and there exist positive constants \(K_0\) ; \(K_1\) such that
$$\begin{aligned} \begin{aligned} \int _Z\left| \gamma (t,u,\phi ,z)\right| _{L^2}^2\lambda (dz)&\le K_0(1+\left| (u,\phi )\right| ^2_{{\mathcal {H}}}),\\ \int _Z\left| \gamma (t,u,\phi ,z)\right| _{L^2}^4\lambda (dz)&\le K_1(1+\left| (u,\phi )\right| ^4_{{\mathcal {H}}}), \\ \left| \sigma _2(t,u,\phi )\right| ^2_{{\mathcal {L}}^2(U;H_2)}&=\sum _{j=1}^\infty \left\| \sigma _j^2(t,u,\phi )\right\| ^2\le K_0, \end{aligned} \end{aligned}$$(2.38)uniformly in \(t \in [0,T ]\) for all \((u,\phi )\in {\mathcal {H}}\).
- (H4):
-
For all \(t\in [0,T]\), there exists a positive constant L such that
$$\begin{aligned}&\sum _{i=1}^{2}\left| \sigma _i(t,u_1,\phi _1)-\sigma _i(t,u_2,\phi _2)\right| ^2_{{\mathcal {L}}^2(U;H_i)}+\int _Z\left| \gamma (t,u_1,\phi _1,z)-\gamma (t,u_2,\phi _2,z)\right| _{L^2}^2\lambda (dz)\nonumber \\&\qquad \le L\left| (u_1,\phi _1)-(u_2,\phi _2)\right| _{{\mathcal {H}}}^2, \end{aligned}$$(2.39)for all \((u_i,\phi _i)\in {\mathcal {H}}\), \(i=1,2\).
Remark 2.3
Condition (2.38)\(_3\) on the noise is widely employed in literature (see [16]). This condition implies that \(\sigma _2(.,0,0)\in L^p(\Omega ,{\mathcal {F}},{\mathbb {P}};L^2(0,T;{\mathcal {L}}^2(U;H_2)))\), for all \(p\ge 2\). Indeed,
For any \((v,\psi )\in {\mathcal {H}},\) we set
where \(c_1>0\) is a constant large enough and independent on \((v,\psi )\) such that \({\mathcal {E}}(v,\psi )\) is non-negative (note that \(F_0\) is bounded from below).
2.3 Abstract Formulation
Using the notations above, we rewrite problem (1.1)–(1.2) as:
which is equivalent to for all \((v,\psi )\in V_1\times H^1(D)\),
\({\mathbb {P}}\)-a.s. and for all \(t\in [0,T]\), for some fixed point \((u_0,\phi _0)\) in \({\mathcal {H}}\).
Remark 2.4
In the weak formulation (2.41), the term \(\mu \nabla \phi \) is replaced by \( A_{1}\phi \nabla \phi \). This is justified since \(f(\phi )\nabla \phi \) is the gradient of \(F(\phi )\) and can be incorporated into the pressure gradient, see [18] for details.
Let us now give the definition of a unique global strong solution in the probabilistic sense to the system (2.41).
Definition 2.1
(Global strong solution) Let the \({\mathcal {F}}_0\)-measurable initial data \((u_0,\phi _0)\in L^4(\Omega ,{\mathcal {F}},{\mathbb {P}};{\mathcal {H}})\) be given. An \({\mathcal {H}}\)-valued \({\mathcal {F}}_t\)-adapted càdlàg process \((u,\phi )(\cdot )\) is called a strong solution to (2.41) if \((u,\phi )\in L^p(\Omega ,{\mathcal {F}},{\mathbb {P}}; L^\infty (0,T;{\mathcal {H}}))\cap L^p(\Omega ,{\mathcal {F}},{\mathbb {P}}; L^2(0,T;{\mathcal {U}})),\) for all \(p\ge 2\) and satisfies (2.42).
Definition 2.2
A strong solution \((u,\phi )(\cdot )\) to (2.41) is called a unique strong solution if \(({\tilde{u}},{\tilde{\phi }})(\cdot )\) is an another strong solution, then
3 Existence and Uniqueness
In this section, we establish the global solvability of the system (2.41). To simplify the notations, throughout this section, we will set (without loss of generality) \(\nu _1=\nu _2=\alpha =\varepsilon ={\mathcal {K}}=1\). We first prove the following energy type equality.
Proposition 3.1
If \((u, \phi )\) is a variational solution to (2.41), then \((u,\phi )\) satisfies
where
Proof
We apply infinite dimensional Itô’s formula (see [33]) to the process \(\left| u\right| _{L^2}^2\) to find
We want now to write Itô’s formula for the free energy functional \({\mathcal {E}}_0(\phi )\), \(\phi \in D(A_1)\). To this end, we should first prove that \({\mathcal {E}}_0: D(A_1)\rightarrow [0,\infty )\) is twice Fréchet differentiable. Let \(\phi ,\psi \in D(A_1)\), using the Taylor-Lagrange formula, we derive that
Owing to the condition (2.37) and the fact that \(D(A_1)\) is continuously embedded in \(H_2\) and in \(L^\infty ({\mathcal {M}})\), we infer that
Therefore,
This proves that the first Fréchet derivative \({\mathcal {D}} : D(A_1) \rightarrow {\mathcal {L}}(D(A_1); {\mathbb {R}})\) of \({\mathcal {E}}_0\) is given by
Also, it is easy to see that \({\mathcal {D}}{\mathcal {E}}_0\) is Fréchet-differentiable with \({\mathcal {D}}^2{\mathcal {E}}_0 : D(A_1) \rightarrow {\mathcal {L}}(D(A_1); {\mathcal {L}}(D(A_1); {\mathbb {R}}))\) given by
Indeed, by direct computation, we can check that
By the embedding of \(D(A_1)\) in \(L^\infty ({\mathcal {M}})\) and in \(L_0^2({\mathcal {M}})\), the mean value theorem and (2.37), we note that
From (3.5), we arrive at
from which we get (3.4). Furthermore, from the hypothesis (H1), we can easily check that the derivatives \({\mathcal {D}}{\mathcal {E}}_0\) and \({\mathcal {D}}^2{\mathcal {E}}_0\) are continuous and bounded on bounded subsets of \(D(A_1)\). Hence, observing that \({\mathcal {D}}^2{\mathcal {E}}_0(\phi )=\mu \), we can apply Itô’s formula to \({\mathcal {E}}_0(\phi )\) in the classical version of [8] to derive that
Adding (3.2) with (3.6) and using (2.8) and the fact that \((u\cdot \nabla \phi , f(\phi ))=0\), we obtain (3.1) which ends the proof of the Proposition 3.1. \(\square \)
Theorem 3.1
We suppose that the Assumptions (H1)–(H4) hold. Moreover, we assume that \(\sigma _1(\cdot ,0,0)\in L^p(\Omega ,{\mathcal {F}},{\mathbb {P}};L^2(0,T;{\mathcal {L}}^2(U;H_1)))\) and that \((u_0,\phi _0)\in L^p(\Omega ,{\mathcal {F}},{\mathbb {P}},{\mathbb {H}})\) satisfies \({\mathbb {E}}{\mathcal {E}}^p(u_0,\phi _0)<\infty \), for all \(p\ge 2\). Then the system (2.41) has a unique strong solution.
The rest of this section is devoted to the proof of Theorem 3.1. The method relies on Galerkin approximation and deterministic Gronwall’s lemma. For the existence part, instead of the Minty-Browder technique used in [27], we prove the existence and certain uniform estimates for the sequence \((u_m,\phi _m)_m\) of the approximation. Then, as in [37], we use the properties of stopping times and some basic convergence principles from functional analysis to prove the existence of the solution.
3.1 Existence of Strong Solution
Let us consider a finite dimensional Galerkin approximation of the system (2.41). Consider \(\{ (w_j,\psi _j),\ j=1,\ldots \}\subset {\mathbb {V}}\) be a orthogonal basis of \({\mathcal {H}}\), where \(\{ w_j,\ j=1,\ldots \}\) and \(\{ \psi _j,\ j=1,\ldots \}\) are eigenvectors of \(A_0\) and \(A_1\) respectively given in the previous section. We set for \(m\in {\mathbb {N}}\), \({\mathcal {H}}_m={\text {span}}\{ (w_1,\psi _1),\ldots .,(w_m,\psi _m) \}=H_{1m}\times H_{2m}\). We look for \((u_m,\phi _m)\in {\mathcal {H}}_m\) solutions to the ordinary differential equations
where \({\mathcal {P}}_m=({\mathcal {P}}_m^1,{\mathcal {P}}_m^2):H_1\times L^2({\mathcal {M}})\rightarrow {\mathcal {H}}_m\) is the orthogonal projection, \(W_m^i(t)={\mathcal {P}}_m^iW^i_t\), for \(i=1,2\). Since the deterministic terms of (3.7) are locally Lipschitz (see Lemma 2.1), and \({\mathcal {P}}_m^i\sigma _i(.)\), \(i=1,2\) and \({\mathcal {P}}_m^1\gamma (.)\) is globally Lipschitz, the system (3.7) has a unique \({\mathcal {H}}_m\)-valued càdlàg local strong solution \((u_m,\phi _m)\in L^2(\Omega ,{\mathcal {F}},{\mathbb {P}};L^\infty (0,T;{\mathcal {H}}_m) )\) with paths \(u\in {\mathbf {D}}(0,T;H_{1m})\) and \(\phi \in C([0; T];H_{2m} )\), \({\mathbb {P}}\)-a.s. (see [1, 27]). Let us now derive the a-priori energy estimates satisfied by the system (3.7).
Proposition 3.2
Let \((u_m,\phi _m)\) be the unique solution of the system (3.7) with \((u_0,\phi _0)\in L^p(\Omega ,{\mathcal {F}},{\mathbb {P}}; {\mathcal {H}})\), for all \(p\ge 2\). If \((u_0,\phi _0)\) is such that \({\mathbb {E}}{\mathcal {E}}^p(u_0,\phi _0)<\infty \) for all \(p\ge 2\), then there exists a positive constant C independent of m such that for all \(p\ge 2\),
Proof
By finite dimensional Itô’s formula (see [4, Theorem 4.4.7]) and the fact that \(b_0(u_m,u_m,u_m)=0\), we obtain for all \(t \in [0, T ]\),
where
Note that (2.30) easily gives
Therefore, using the fact that \(\left| x\right| ^2_{L^2}-\left| y\right| ^2_{L^2}+\left| x-y\right| ^2_{L^2}=2(x-y,x),\) \(\forall x, y\in H_1\), we infer from (3.10) that
Note that, since \({\text {spam}}\{ \psi _1,\ldots .\psi _m \}\subset D(A_1)\), we infer that \(D{\mathcal {E}}_0(\phi _m)=\mu _m\). Therefore, applying Itô’s formula to the process \({\mathcal {E}}_0(\phi _m)\), we get
Using the fact that \(H_1\hookrightarrow L^p({\mathcal {M}})\), \(p\ge 2\), by assumption (H1), we get
Now, for each \(n\ge 1,\) we consider the \({\mathcal {F}}_t\)-stopping time \(\tau _n^m\) defined by:
For fixed m, the sequence \(\{\tau _n^m;n\ge 1 \}\) is increasing to T. Adding (3.10) with (3.14) after using (3.12) and (3.15), we get for all \(t\in [0,T]\),
Now raising both sides to the power \(p\ge 2\), taking supremum over \(s \in [0, t\wedge \tau _n^m]\) and taking mathematical expectation we have
where
It follows that
Applying Burkholder–Davis–Gundy’s inequality (see [33, Theorem 48]) and using (3.20)–(3.21), we derive that
Using Hölder’s inequality, it follows from (3.19) that
By Doob’s inequality, we derive will the help of (2.38)\(_1\) and Hölder’s inequality that
By Hölder’s inequality, we derive that
By (2.37), (3.7)\(_4\), and the fact that \(H_1\hookrightarrow L^2({\mathcal {M}})\) we infer that
By Doob’s inequality, Hölder’s inequality, the condition (2.29), (3.26) and Young’s and Poincaré-Wirtinger’s inequalities, we infer that
It follows from (3.18)–(3.27) that
(3.8) follows from Gronwall’s lemma and the fact that \(\tau _n^m\nearrow T\) as n goes to \(\infty \).
By Young’s inequality, and (3.8), we derive that
From (3.29) and (3.30) we get (3.9). The Proposition 3.2 is then proved. \(\square \)
Corollary 3.1
Under the same hypothesis as in Proposition 3.2, there exists a positive constant C independent of m such that for all \(p\ge 2 \),
Proof
By (3.7)\(_4\), (2.37) and the Poincaré–Wirtinger inequality, we note that
The estimates (3.31), (3.32) and the first part of (3.33) follows from (3.36)–(3.38) and the Proposition 3.2. By (2.10), (2.11) and (2.12), we also note that
By (3.39)–(3.41) we end the proof of Corollary 3.1. \(\square \)
From the Corollary 3.1, and along with the Banach-Alaoglu theorem, one can extract a subsequence still denoted by \((u_m,\phi _m)\) to simplify the notation which converges to the following limits
With these convergence at hand we see from (3.7) that \((u,\phi )(.)\) satisfies the following Itô stochastic differential: For all \(t\in [0,T]\),
\({\mathbb {P}}\)-almost surely as equality in \(V_1^*\times H_2^*\).
From the energy estimates (see Corollary 3.1), \((u_m,\phi _m)\) is almost surely uniformly convergent on finite intervals [0, T] to \((u,\phi )\), from which it follows that \((u,\phi )\) is \({\mathcal {F}}_t\)-adapted and the \({\mathcal {F}}_t\)-adapted paths of u are càdlàg while the \({\mathcal {F}}_t\)-adapted paths of \(\phi \) are continuous (see [4, Theorem 6.2.3]).
Proposition 3.3
We have the following identities
Proof
Let \(({\tilde{u}}_m,{\tilde{\phi }}_m,{\tilde{\mu }}_m)={\mathcal {P}}^0_m(u,\phi ,\mu ),\) where \({\mathcal {P}}^0_m=({\mathcal {P}}^1_m,{\mathcal {P}}^2_m, {\mathcal {P}}^2_m)\). We have
From (3.7) and (3.43), we derive that for \( 1\le k\le m\),
Note that since \(B_0,\) \(R_0\) and \(B_1\) are bilinear, we derive that
Let us set \(\theta _m={\tilde{u}}_m-u_m,\) \(\rho _m={\tilde{\phi }}_m-\phi _m,\) \(\zeta _m={\tilde{\mu }}_m-\mu _m.\) From Itô’s formula, we have
where
It follows that
Also, applying the Itô formula to the process \(\left\| \rho _m\right\| ^2\), and replacing \(\psi _k\) in (3.47)\(_3\) by \({\overline{\zeta }}_m-\xi \rho _m\), we obtain
Note that, owing to \(\langle B_0(u_m,\theta _m),\theta _m \rangle =b_0(u_m,\theta _m,\theta _m)=0,\) we have
Also we have
Recall that from [18], there exists a monotone non-decreasing function \(Q_1(x_1,x_2)\) such that
where
Let us set
where \(\xi \) is small enough such that \(1-c\xi >0.\) Also, let us set
Adding (3.50) with (3.51), using (3.52)–(3.59), it follows from Itô’s formula that
Now, for each \(n\ge 1,\) we consider the \({\mathcal {F}}_t\)-stopping time \(\tau _n\) defined by:
We derive from (3.61) that
Now, we want to prove that the right side of (3.62) goes to 0 as m goes to \(+\infty .\) We first note that, since \(0<\sigma (t)\le 1\) and \( ({\tilde{u}}_m,{\tilde{\phi }}_m)\longrightarrow (u,\phi )\) in \(L^2(\Omega ,{\mathcal {F}},{\mathbb {P}};L^{2}(0,T;{\mathbb {V}}))\), we have
Following the same way as in [14, 36], we derive that
Since \({\mathcal {P}}_m^i\circ {\mathcal {P}}_m^i={\mathcal {P}}_m^i\), and \(\left\| {\mathcal {P}}_m^i\right\| \le 1\), \(i=1,2,\) it follows that
Therefore, as \({\mathcal {P}}_m^1\gamma (\cdot ,u_m(\cdot ),\phi _m(\cdot ))\rightharpoonup \Psi (.)\) in \({\mathcal {M}}^2_T(H_1)\) and \({\mathcal {P}}_m^i\sigma _i(\cdot ,u_m,\phi _m)\longrightarrow \varPhi _i(\cdot )\) in \( L^2(\Omega ,{\mathcal {F}},{\mathbb {P}};L^{2}(0,T;{\mathcal {L}}^2(U;H_i))),\) \(i=1,2,\) we see that
This concludes that the right side of (3.62) goes to 0 as m goes to \(+\infty .\)
Now using the fact that \(1_{[0,\tau _n]}\sigma (t)\le 1,\) we derive from (3.62) that
We note that from (3.66)\(_{3,4,5}\) and the fact that the sequence \(\{\tau _n;\ n \ge 1\}\) is increasing to T, we derive that
The end of the proof of the Proposition 3.3 is very similar to [36, Proof of Claim 2]. \(\square \)
By Proposition 3.3, we infer from (3.43) that \((u,\phi )\) is a strong solution of problem (1.1) in the sense of Definition 2.1.
3.2 Uniqueness of Strong Solution
Assume that \((u_1,\phi _1)\) and \((u_2,\phi _2)\) are two strong solutions to (1.1). We set \((w,\psi ,\mu )=(u_1,\phi _1,\mu _1)-(u_2,\phi _2,\mu _2)\), \({\tilde{\sigma }}_i(\cdot \cdot )=\sigma _i(\cdot ,u_1(\cdot ),\phi _1(\cdot ))-\sigma _i(\cdot ,u_2(\cdot ),\phi _2(\cdot ))\), \(i=1,2\) and \({\tilde{\gamma }}(\cdot ,\cdot )=\gamma (\cdot ,u_1(\cdot ),\phi _1(\cdot ),\cdot )-\gamma (\cdot ,u_2(\cdot ),\phi _2(\cdot ),\cdot )\). Then \((w,\psi )\) satisfies the following system
We apply infinite dimensional Itô’s formula (see [33]) to the process \(\left| w\right| ^2_{L^2}\) and using the fact that \(\left| x\right| ^2_{L^2}-\left| y\right| ^2_{L^2}+\left| x-y\right| ^2_{L^2}=2(x-y,x),\) \(\forall x, y\in H_1\) to find that
Also, applying the Itô formula to the process \(\left\| \psi \right\| ^2\), we get
Now we take the duality of (3.67)\(_3\) with \(A_1(\mu -\langle \mu \rangle ) -\xi A_1\psi \), where \(\xi > 0\) is small enough and will be selected later. Adding the resulting equality to (3.68) and (3.69), we derive that
We note that
where \(Q_1\) is a suitable monotone non-decreasing function independent on time and the initial condition.
Now, let us set \({\mathcal {Y}}_2(t)=\left| w(t)\right| ^2_{L^2}+\left\| \psi \right\| ^2,\) and
So, applying Itô’s formula, to the real-valued process \(\sigma (t){\mathcal {Y}}_2(t)\), using (3.70) and the inequalities (3.71)–(3.77), we derive that
Note that the expectation of the stochastic integrals in (3.70) varnishes. Therefore we obtain
It follows from the deterministic Gronwall lemma that \({\mathcal {Y}}_2(t)=0\) \({\mathbb {P}}\)-a.s., for all \(t\in [0,T].\) Hence \((u_1,\phi _1)=(u_2,\phi _2),\) \({\mathbb {P}}\)-a.s., for all \(t\in [0,T].\) Note that in (3.70), we choose \(\xi >0\) and small enough such that \(1-c\xi >0.\)
4 Exponential Behavior
In this section, we show some aspects of the effects produced in the long-time behavior of the solution to a two dimensional Cahn–Hilliard–Oldroyd model with order one for the non-Newtonian two phase fluid flows under the presence of stochastic perturbations. More precisely, we discuss the moment exponential stability and almost sure exponential stability of strong solutions \((u,\phi )\) of stochastic 2D Cahn–Hilliard–Oldroyd model under some conditions.
We will consider the following system
where \(g=(g_1,g_2):[0,T]\rightarrow V_1^*\times H_2\) is Borel measurable function such that \( g\in L^2(0,T;V_1^*\times H_2)\).
Remark 4.1
From the previous section, it is clear that for \( g=(g_1,g_2)\in L^2(0,T;V_1^*\times H_2)\), there exists a unique (pathwise) global strong solution for the system (4.1) under the hypothesis (H1)–(H4).
Hereafter, as in [37], we assume that f satisfies the additional condition. For all \(\phi _1,\phi _2\in D(A_1^{3/2})\),
where \(\alpha _0 > 0\) is a positive constants independent of \(\phi _1\) and \(\phi _2\).
Assuming that g is independent of t, we now consider the following stationary equation
Then we recall the following solvability result for the system (4.4) for \(\nu =(\nu _1+ \frac{\gamma }{\delta } )>0\), where the proof is very similar to [37, Section 3.1].
Lemma 4.1
If \( g=(g_1,g_2)\in V_1^*\times V_2^*\), then there exists a stationary solution \((u^*,\phi ^*)\in {\mathcal {U}}\) to system (4.4), Moreover, for \(\varepsilon >0\) large enough such that \(\alpha _2=\min ({\mathcal {K}}^{-1}\nu , \varepsilon ^2\nu _2-\varepsilon \alpha _0)\) is non negative, if \(\alpha _2-2(\left\| g_1\right\| ^2_{V_1^*}+\left\| g_2\right\| _{V_2^*})>0\), then the stationary solution to (4.4) is unique.
Now, we give the definition of exponential stability.
Definition 4.1
We say that a strong solution \((u, \phi )(t)\) to (4.1) converges to \((u^*, \phi ^*)\in {\mathcal {H}}\) exponentially in the mean square if there exists \(a>0\) and \(M_0 = M_0((u, \phi )(0)) > 0\) such that
If \((u^*, \phi ^*)\) is a solution to (4.4), we say that \((u^*, \phi ^*)\) is exponentially stable in the mean square provided that every strong solution to (4.1) converges to \((u^*, \phi ^*)\) exponentially in the mean square with the same exponential order \(a > 0.\)
Theorem 4.1
Let \((u^*,\phi ^*)\) be the unique stationary solution of (4.4) and \(\sigma _i(s,u^*,\phi ^*)=0\), \(i=1,2\), \(\gamma (s,u^*,\phi ^*,z)\), for all \(s>0\) and \(z\in Z\). Suppose that the assumption (H1)–(H5) are satisfied, then the strong solution \((u,\phi )(t)\) of system (4.1) converges to the stationary solution \((u^*,\phi ^*)\) of the system (4.4) is exponentially stable in the mean square provided that \(\varepsilon \) is large enough such that \(\alpha _2>0\) and the following inequality holds
where \(\alpha _2=\min ({\mathcal {K}}^{-1}\nu _1,\nu _2\varepsilon ^2-\alpha _0\varepsilon )\), \(\alpha _3=\max ({\mathcal {K}}^{-1},\varepsilon )\) and \(c_1>0\) is given below.
Proof
With the condition (4.6), one can chose a constant \(a > 0\) such that
We set \((w,\psi )(t)=(u,\phi )(t)-(u^*,\phi ^*).\)
Applying the infinite dimensional Itô formula to the process \({\mathcal {K}}^{-1}e^{2a t}\left| w(t)\right| ^2_{L^2}\) we get
Applying the Itô formula to the process \(\varepsilon e^{2a t}\left| \nabla \psi (t)\right| ^2_{L^2},\) we derive that
Summing (4.8) and (4.9), after using (4.3), we derive that
Using the definition of \(\beta =\gamma e^{-\delta t}\), we note that
Using (4.11), we infer from (4.4) that \((u^*,\phi ^*)\) satisfies
Using (4.10) and (4.12), we derive that
Using (2.30), we have
Using Cauchy-Schwarz, Hölder’s and Young’s inequalities, we get
for \(0<a<\delta \).
Note that
Using (2.39), (4.16)–(4.19) and (2.23) in (4.13) we get
Since a satisfies (4.7), we finally have
and hence \((u,\phi )(t)\) converges to \((u^*,\phi ^*)\) exponentially in the mean square. \(\square \)
Theorem 4.2
Suppose that all conditions given in Theorem 4.1 are satisfied, then the strong solution \((u, \phi )(t)\) of (4.1) converges to the stationary solution \((u^*, \phi ^*)\) of (4.4) almost surely exponentially.
Proof
Let \(n=1,2,\ldots .,\) and \(h>0\). By the Itô formula, for any \(t \ge N\) we have
Taking supremum from nh to \((n+1)h\) and then taking expectation in (4.22) after using (4.14)–(4.19) with \(a=1\), we find
By, Davis’, Hölder’s, and Young’s inequalities we derive
An application of the Burkholder–Davis–Gundy inequality (see [33, Theorem 48]), Hölder’s, and Young’s inequalities yield
Combining (4.24) and (4.25), substituting it in (4.23), and then using (2.39), we get
where
Using (4.21) in (4.26), we arrive at
For \(\theta \in (0,a)\), we set
By Chebychev’s inequality, we also have
which implies that
Therefore, by the Borel-Cantelli lemma, there is a finite integer \(n_0(\omega )\) such that
for all \(n\ge n_0\), and the Theorem 4.2 is then proved. \(\square \)
For the next theorem we assume that \(g_1\) and \(g_2\) depend on \(u(\cdot )\), \(\phi (\cdot )\) and satisfy the following Lipschitz condition: For all \((u_1,\phi _1), (u_2,\phi _2)\in {\mathcal {U}}\),
Theorem 4.3
If \(g_i(0,0)=0\), \(\sigma _i(t,0,0)=0\), \(i=1,2\) and \(\gamma (t,0,0,z)=0\), for all \(t> 0\) and \(z\in Z\), then any strong solution \((u,\phi )(t)\) to (4.1) converges to zero almost surely exponentially if
Proof
Owing to (4.31), one can chose a constant \(a > 0\) such that
Applying the infinite dimensional Itô formula to the process \({\mathcal {K}}^{-1}e^{2a t}\left| u(t)\right| ^2_{L^2}\) and \(\varepsilon e^{2a t}\left| \nabla \psi (t)\right| ^2_{L^2}\) respectively, summing the results and using (2.7)–(2.8), we get
Using (2.30), (4.3), (2.39) and (4.30), we infer from (4.33) that
This implies that
under the condition (4.31), it is immediate that
This implies that the strong solution of (4.1) converges to zero exponentially in the mean square. We can then finish the proof using the same method as in the proof of Theorem 4.2. \(\square \)
References
Albeverio, S., Brzeźniak, Z., Wu, J.L.: Existence of global solutions and invariant measures for stochastic differential equations driven by Poisson type noise with non Lipschitz coefficients. J. Math. Anal. Appl. 371(1), 309–322 (2010)
Agranovich, Y.Y., Sobolevskii, P.E.: Investigation of viscoelastic fluid mathematical model. RAC. Ukran. SSR Ser. A 10, 71–73 (1989)
Anderson, D.M., McFadden, G.B., Wheeler, A.A.: Diffuse-interface methods in fluid mechanics. Annu. Rev. Fluid Mech. 30, 139–165 (1998)
Applebaum, D.: Lévy Processes and Stochastic Calculus. Cambridge Studies in Advanced Mathematics. Cambridge University Press, Cambridge (2004)
Barbu, V.: Nonlinear Semigroups and Differential Equations in Banach Spaces. Noordhoff, Leyden (1976)
Brzeźniak, Z., Hausenblas, E., Razafimandimby, P.A.: Some results on the penalised nematic liquid crystals driven by multiplicative noise, arXiv:1310.8641 (2016)
Brzeźniak, Z., Manna, U., Panda, A.A.: Existence of weak martingale solution of nematic liquid crystals driven by pure jump noise, arXiv:1706.05056 (2017)
Da Prato, G., Zabczyk, J.: Stochastic Equations in Infinite Dimensions, 2nd edn. Cambridge University Press, Encyclopedia of Mathematics and its Applications (2014)
Deugoué, G., Jidjou Moghomye, B., Tachim Medjo, T.: Existence of a solution to the stochastic nonlocal Cahn–Hilliard Navier–Stokes model via a splitting-up method. Nonlinearity 6, 66 (2020)
Deugoué, G., Jidjou Moghomye, B., Tachim Medjo, T.: Fully discrete finite element approximation of the stochastic Cahn–Hilliard–Navier–Stokes system (submitted)
Deugoué, G., Ndongmo Ngana, A., Tachim Medjo, T.: Martingale solutions to stochastic nonlocal Cahn–Hilliard–Navier–Stokes equations with multiplicative noise of jump type. Phys. D 398, 23–68 (2019)
Deugoué, G., Ndongmo Ngana, A., Tachim Medjo, T.: On the strong solutions for a stochastic 2D nonlocal Cahn–Hilliard–Navier–Stokes model. Dyn. PDE 17(1), 19–60 (2020)
Deugoué, G., Tachim Medjo, T.: Convergence of the solutions of the stochastic 3D globally modified Cahn–Hilliard–Navier–Stokes equations. J. Differ. Equ. 265(2), 545–592 (2018)
Deugoué, G., Tachim Medjo, T.: On a stochastic 2D Cahn–Hilliard–Navier–Stokes system driven by jump noise. Commun. Stoch. Anal. 13(1), 66 (2019)
Doi, M.: Dynamics of domains and textures. In: McLeish, T.C. (Ed.) Theoretical Challenges in the Dynamics of Complex Fluids, NATO-ASI Series, vol. 339, pp. 293–314 (1997)
Feireisl, E., Petcu, M.: Stability of strong solutions for a model of incompressible two-phase flow under thermal fluctuations. J. Differ. Equ. 267(3), 1836–1858 (2019)
Feng, X.: Fully discrete finite element approximations of the Navier–Stokes–Cahn–Hilliard diffuse interface model for two-phase fluid flows. SIAM J. Numer. Anal. 44, 1049–1072 (2006)
Gal, C.G., Grasselli, M.: Asymptotic behavior of a Cahn–Hilliard–Navier–Stokes system in 2D. Ann. Inst. H. Poincaré Anal. Non Linéaire 27(1), 401–436 (2010)
Gurtin, M.E., Polignone, D., Vinals, J.: Two-phase binary fluids and immiscible fluids described by an order parameter. Math. Models Methods Appl. Sci. 6, 8–15 (1996)
Hohenberg, P.C., Halperin, B.I.: Theory of dynamical critical phenomena. Rev. Mod. Phys. 49, 435–479 (1977)
Kay, D., Welford, R.: Efficient numerical solution of Cahn–Hilliard–Navier–Stokes fluids in 2D. SIAM J. Sci. Comput. 29, 2241–2257 (2007)
Kay, D., Styles, V., Welford, R.: Finite element approximation of a Cahn–Hilliard–Navier–Stokes system. Interfaces Free Bound. 10, 15–43 (2008)
Kotsiolis, A.A., Oskolkov, A.P.: On the solvability of fundamental initial-boundary value problem for the motion equations of Oldroyd’s fluid and the behavior of solutions, when \(t\rightarrow \infty \). Notes Sci. LOMI. 150(6), 48–52 (1986)
Li, F., You, B.: Random attractor for the stochastic Cahn–Hilliard–Navier–Stokes system with small additive noise. Stoch. Anal. Appl. 36, 66 (2018)
Manna, U., Mukherjee, D.: Weak solutions and invariant measures of stochastic Oldroyd-B type model driven by jump noise. J. Differ. Equ. 272, 760–818 (2021)
Mohan, M.T.: Well posedness, large deviations and ergodicity of the stochastic 2D Oldroyd model of order one. Stoch. Process. Appl. 130, 4513–4562 (2020)
Mohan, M.T.: Deterministic and stochastic equations of motion arising in Oldroyd fluids of order one: existence, uniqueness, exponential stability and invariant measures. Stoch. Anal. Appl. 38(1), 1–61 (2019)
Mohan, M.T.: On the three dimensional Kelvin–Voigt fluids: global solvability, exponential stability and exact controllability of Galerkin approximations. Evol. Eq. Control Theory 9, 301–339 (2019)
Mohan, M.T., Sritharan, S.S.: Stochastic Navier–Stokes equation perturbed by Lévy noise with hereditary viscosity. Infinite Dimen. Anal. Quantum Probab. Relat. Top. 22(1), 1950006 (2019)
Morro, A.: Phase-field models of Cahn–Hilliard fluids and extra fluxes. Adv. Theor. Appl. Mech. 3, 409–424 (2010)
Oskolkov, P.A.: Initial boundary value problems for the equations of motion of Kelvin–Voigt fluids and Oldroyd fluids. Proc. Steklov Inst. Math. 2, 137–182 (1989)
Ortov, P.V., Sobolevskii, P.E.: On mathematical models of a viscoelastic with a memory. Differ. Integr. Equ. 4, 103–115 (1991)
Protter, P.E.: Stochastic Integration and Differential Equations, 2nd edn. Springer, New York (2005)
Sobolevskii, P.E.: Asymptotics of stable viscoelastic fluid motion. Oldroyd’s mathematical model. Math. Nachr. 177(1), 281–305 (1996)
Temam, R.: Navier–Stokes Equations, Theory and Numerical Analysis. Studies in Mathematics and Its Applications, vol. 2, 3rd edn. North-Holland, Amsterdam (1984)
Tachim Medjo, T.: On the existence and uniqueness of solution to a stochastic 2D Cahn–Hilliard–Navier–Stokes model. J. Differ. Equ. 263(2), 1028–1054 (2017)
Tachim Medjo, T.: The exponential behavior of a stochastic Cahn–Hilliard–Navier–Stokes model with multiplicative noise. Commun. Pure Appl. Anal. 18(3), 1117–1138 (2019)
Tachim Medjo, T.: Weak solution of a stochastic 3D Cahn–Hilliard–Navier–Stokes model driven by jump noise. J. Math. Anal. Appl. 484(1), 123680 (2020)
Acknowledgements
The authors would like to thank the anonymous referees whose comments help to improve the contain of this article.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Additional information
Communicated by F. Flandoli.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Deugoué, G., Jidjou Moghomye, B. & Tachim Medjo, T. Existence and Exponential Behavior for the Stochastic 2D Cahn–Hilliard–Oldroyd Model of Order One. J. Math. Fluid Mech. 24, 15 (2022). https://doi.org/10.1007/s00021-021-00647-2
Accepted:
Published:
DOI: https://doi.org/10.1007/s00021-021-00647-2